Elevated ctDNA Tumor Fraction Is Associated with Improved Mutation Detection but Worse Overall Survival in Advanced Non–Small Cell Lung Cancer: A Lung-MAP Study

Abstract Purpose: ctDNA is a powerful diagnostic companion to tissue profiling. Tumor fraction (TF) is a global assessment of an individual’s ctDNA burden. We evaluated the impact of plasma TF on mutation detection and clinical outcomes in patients with previously treated, advanced non–small cell lung cancer on the Lung Master Protocol (Lung-MAP). Experimental Design: Paired tumor tissue and plasma were collected prospectively from patients on the Lung-MAP study. Plasma was collected within 30 days of a new biopsy with no intervening therapies. Tissue and ctDNA genomic profiling and ctDNA TF levels were assessed by Foundation Medicine. TF was primarily calculated from tumor aneuploidy, defaulting to fragmentomics and maximum somatic allele frequencies when aneuploidy was not detectable. The effect of TF on tissue–plasma mutation concordance, overall survival, and its relation to variant allele frequencies was assessed using linear regression, Lin's coefficient, and Cox modeling/log-rank testing. Results: A total of 194 patients were eligible for analysis. TF ≥1% was significantly associated with improved positive percent agreement between ctDNA and tissue across multiple alteration types with the exception of copy-number gains. For short variants, positive percent agreement improved from 51% when TF <1% to 95% when TF ≥1%. TF showed a significant robust correlation with variant allele frequency for KRAS, STK11, and TP53—the three most common mutations. TF <1% was significantly associated with improved patient overall survival compared with TF ≥1% or TF ≥10%. Conclusions: TF provides an accurate, clinically useful assessment of ctDNA plasma levels from patients with refractory, advanced non–small cell lung cancer. TF levels ≥1% are associated with significantly worse overall survival but improved mutation detection in liquid biopsies.